OpenAI announced that its core API —— Responses API now supports **Model Context Protocol (MCP)**, and has made significant updates to image generation, Code Interpreter, and file search tools. These upgrades greatly simplify the development process of AI agents, allowing developers to connect AI agents to external tools and services with just a few lines of code, further enhancing the flexibility and functionality of the API.
MCP Support: Simplifying AI Agent Development
OpenAI’s addition of MCP support through the Responses API marks another major breakthrough in the field of AI agent development. MCP (Model Context Protocol) is a protocol for connecting AI models to external tools and services. Developers can achieve complex function integration with just a few lines of code. For example:
- You can connect an AI agent to the e-commerce platform **Shopify** with **9 lines of code**, enabling seamless interaction with the e-commerce system.
- You can connect an AI agent to the cloud communication platform **Twilio** with **13 lines of code**, supporting real-time communication functions.
OpenAI also joined the MCP Steering Committee to promote the development of this ecosystem, providing developers with more standardized tool integration solutions.
Image Generation: Real-Time Preview and Multi-Round Editing
The Responses API has integrated OpenAI’s latest image generation model **gpt-image-1**, providing developers with enhanced visual creation capabilities. The new features include:
- **Real-Time Preview**: Developers can view the generation results in real time during the image generation process.
- **Multi-Round Editing**: Supports step-by-step adjustments and optimizations of images, allowing developers to make fine-grained modifications according to their needs.
This update is based on the popular GPT-4o model, whose image generation capability once sparked a "Studio Ghibli" style anime craze online, showcasing its diverse style generation capabilities.
Code Interpreter: Data Analysis and Image Processing
The upgraded **Code Interpreter** tool further enhances the practicality of the Responses API, enabling AI models to:
- Perform complex data analysis tasks.
- Solve advanced mathematical problems.
- Process and manipulate image data, such as generating or editing images through code.
This feature provides developers with powerful tools applicable to various scenarios, from financial analysis to software development.
File Search: Efficient Retrieval Across Multiple Data Sources
The improvements to the file search tool allow AI agents to extract relevant information from documents more efficiently. The updated features include:
- Extracting relevant content blocks from documents based on user queries.
- Supporting searches across multiple data sources, enhancing the breadth and precision of information retrieval.
This upgrade provides stronger support for applications in market research, education, and software development.
OpenAI stated that the pricing for the new features remains consistent with existing rates. For example:
- Code Interpreter is charged at $0.03 per session.
- File search is charged at $2.50 per 1,000 calls, with storage costs at $0.10 per GB per day (the first GB is free).
- Image generation is charged based on resolution and quality levels, starting at $0.011 per image.
All features went live on May 21, 2025, and developers can find detailed implementation guidelines in the official OpenAI documentation.
Summary
OpenAI’s update to the Responses API, by supporting MCP, enhancing image generation, Code Interpreter, and file search functions, provides developers with a more powerful and flexible toolkit. Whether connecting external services like Shopify and Twilio or performing complex data analysis and image processing, the Responses API demonstrates its great potential in AI agent development. Developers can now implement more complex AI applications with fewer lines of code, marking a significant step forward for OpenAI in advancing the application of AI technology.
OpenAI Developer Platform: https://platform.openai.com